From traffic routing to self-driving cars, Alexa to Siri, AI’s reach is extending into all areas of life, including healthcare. Join Kimberley to learn more about how AI is being used now, and will be used in the near future, to facilitate provider-patient communication, mine medical records, assess patients, predict illness, suggest treatments, and so much more. This class is freshly updated for 2023 and also includes a section on the bias inherent in AI, which impacts the kind of treatment that patients receive.
2. Learning Objectives
• Learn to define “Artificial Intelligence”
• Learn about the history of AI
• Learn the difference between AI and machine learning
• Learn about AI industry leaders
• Learn about popular current uses of machine learning in daily life
• Learn about current healthcare applications for AI
• Learn about barriers to AI
• Learn about possible AI uses in the near future
3. AI: definition
•“…the theory and development of computer
systems able to perform tasks that normally
require human intelligence, such as visual
perception, speech recognition, decision-
making, and translation between languages.”
4. AI: a brief history
•An ancient idea
•Talos & Galatea
5. AI: a brief history, 2
• 1206- Al-Jazari creates a programmable orchestra of
mechanical human beings
• 1580- Rabbai Judah Loew Ben Bezalel (allegedly) creates
the Golem
• 1642- Blaise Pascal invents the first digital calculating
machine
• 1726- Jonathon Swift publishes Gulliver’s Travels à the
Engine:“a Project for improving speculative Knowledge by
practical and mechanical Operations ”
6. AI: a brief history 3
• 1818- Mary Shelley publishes Frankenstein; or The Modern Prometheus-
speculates on the ethics of creating life
• 1863- Samuel Butler speculates that machines will one day become
conscious and supplant humanity
• 1941- Konrad Zuse builds first working program-controlled computer
• 1945- Game theory is introduced in Theory of Games and Economic
Behavior; integral to development of modern AI
• 1950- Alan Turing introduces the idea of the Turing Test
• 1950- Isaac Asimov publishes The Three Laws of Robotics
7. AI: a brief history 4
• 1951- first working AI programs written; checkers and chess
• 1955- Arthur Samuels builds a program which learns how to play checkers
• 1956- Dartmouth College Summer AI Conference organized (term
“artificial intelligence” is coined)
• 1959- Jonathon McCarthy and Marvin Minsky found the MIT AI Lab
• Late 1950’s-early 1960’s- Margaret Masterson and colleagues design
semantic nets for machine translation
• 1960’s- Ray Solomonoff lays foundation of mathematical theory of AI
8. AI: a brief history, 5
• 1963- ANALOGY (written by Thomas Evans) demonstrates that computers
can solve the same analogy questions that are given on IQ tests
• 1965- ELIZA- interactive program that carries on a dialogue in English, on any
topic
• 1965- DENDRAI, 10-year effort to to deduce the molecular structure of
organic compounds using scientific instrument data. First expert system
• 1966- Ross Quillian demonstrates semantic nets
• 1969- Shakey the Robot demonstrated animal locomotion, perception and
problem-solving
9. AI: a brief history 6
• Early 1970’s- Jane Robinson establishes a Natural Language
Processing Center
• 1973- Assembly Robotics Group builds Freddy Robot; capable of
using visual perception to locate and assemble models
• 1975- Marvin Minsky publishes “Frames”; brings together ideas
about schemas and semantic links
• 1979-the Stanford Cart; first computer-controlled automated
vehicle. Successfully navigated a room full of chairs
• 1986- robot cars from the University of Munich drove up to 55 mph
on empty streets
10. AI: a brief history 7
• 1986- Barbara Grosz and Candace Sidner create the first computation model
of discourse, establishing the field of research.
• 1997- Deep Blue chess machine defeats the (then) world chess champion,
Garry Kasparov.
• 1998- Furby released (first successful attempt at producing a type of A.I to
reach domestic market)
• 1998- Tim Berners-Lee publishes Semantic Web Road Map
• Late 1990’s- Web crawlers and other AI-based information extraction
programs become essential in widespread use of the Internet
11. AI: a brief history 8
• 2000- Nomad robot explores remote regions of Antarctica looking for
meteorite samples
• 2002- Roomba released (autonomous vacuum)
• 2004- DARPA Grand Challenge (prize money for autonomous vehicles)
• 2004- “Spirit” and “Opportunity” autonomously navigate Mars
• 2005- Recommendation technology based on tracking web activity brings
AI to marketing.
• 2005- Blue Brain- project to simulate the brain at molecular detail
• 2009- Google self-driving car
12. AI: a brief history 9
• 2010- Microsoft Kinect (machine learning for human motion capture)
• 2011- Watson defeats Jeopardy! champions
• 2011-2014- Siri, Google Now, Cortana (natural language; recommendations;
perform actions)
• 2013- NEIL (Never Ending Image Learner) released at Carnegie Mellon
University; constantly compared and analyzed relationships between
different images
• 2015- Hawking, Musk, Wozniak and 3,000 researchers in AI and robotics sign
open letter calling for ban on research of autonomous weapons
13. AI: a brief history 10
• 2017- Asilomar Conference on Beneficial AI; discussed AI ethics and
strategies for bringing about beneficial AI while avoiding risk from artificial
general intelligence.
• 2018-
• Google’s Duplex impressed a trade show audience with its human-
sounding voice and ability to make calls on behalf of clients
• Amazon’s Rekognition facial recognition software caused concern due to
ethical concerns
• Google’s Project Maven was a collaboration with the Department of
Defense to allow the use of its API in drones
14. AI: a brief history, 11
• 2019
• FDNA uses AI to detect rare diseases
• “Genomic insights through next-generation phenotyping (NGP)”
• 2020
• OpenAI announced DALL-E
• multimodal AI system that generates images from text
• 2022
• Midjourney , an artificial intelligence art generation service, enters open beta
• Chatgpt launched by OpenAI
15. Examples of AI-Generated Avatars
These were created with different AI generators and needed between 8-20 pictures of me to generate
the images.
16. According to pwc’s
research, AI’s estimated
potential contribution to
the global economy by 2030
will be $15.7 trillion.
17. The Interactive Data
Explorer from pwc:
https://www.pwc.com/gx/en/i
ssues/data-and-
analytics/publications/artificial
-intelligence-
study.html#explorer
18. Neural Networks- crucial to development
•“A Neural Network is a computer system
designed to work by classifying
information in the same way a human brain
does. It can be taught to recognize, for
example, images, and classify them
according to elements they contain.”
20. AI in (Old) Pop Culture
• C-3PO
• Skynet
• Baymax
• WOTAN (Doctor Who)
• Omnius (Dune)
• Cylons
• Transformers
• The Matrix
21. AI vs. Machine Learning
• AI is the broad category;
machine learning is one application of AI
• “AI is basically the intelligence –
how we make machines intelligent,
while machine learning is the
implementation of the computer methods
that support it.”
For examples, if the category
was pasta, tortellini would be
one type of pasta.
22. Machine Learning- definition
• “Machine learning is an application of artificial intelligence (AI) that
provides systems the ability to automatically learn and improve from
experience without being explicitly programmed. Machine learning
focuses on the development of computer programs that can access
data and use it learn for themselves.”
• *pattern recognition
• ML is really “start of the art” of AI; true AI isn’t a reality yet
23. Examples of Machine Learning
• Virtual Personal Assistants- Siri, Alexis, Cortana
• Traffic predictions
• Surveillance systems
• Computer vision à how Pinterest knows
which pins to recommend
• Facial recognition
• Chatbots- Chatgpt, FB, Replika, Dadbot
• Shopping recommendations (Alibaba)
24. Examples of Machine Learning
• Purchase predictions
• Video games
• Self-driving cars
• Fraud detection
• News generation
• Pandora
27. Let’s talk about Chatgpt…
• What is it?
• Chat Generative Pre-Trained Transformer is a Large Language Model (LLM)
machine learning chatbot, created by AINOW and released in November 2022
• Why all the excitement?
• Chatgpt is the first chatbot of its kind to be able to write and CONVERSE
convincingly in English.
• How is it being used in the sciences?
• Write essays & talks
• Summarize literature
• Improve papers
• Identify research gaps
• Write computer code
• Perform statistical analyses
28. Chatgpt- on the positive side…
• How might it revolutionize research practices?
• Accelerate the innovation process
• Shorten time-to-publication
• Make science more equitable
• Increase the diversity of science perspectives
- From “ChatGPT: five priorities for research”
29. Chatgpt, on the negative side…
• Gets things wrong, spurring misinformation- and it’s very convincing
• Needs to be carefully monitored/edited by a human subject matter
expert
• Its writings lack depth
• It lacks its own set of ethics, relying on the sources from whom it learns
to “create” them
30. Chatgpt: an example from a friend:
“write an incident report for an RA conducting
rounds who identified a broken window”
31.
32.
33. Examples of AI start-ups
• Zipline- blood & vaccine delivery via drones
• Everlaw- trial preparation through document analysis
• Voicera- personal assistant
• ShieldAI- drones in combat situations (mapping,
identification)
34. Why AI in healthcare?
•Save time/efficiency
•Shortage of clinicians
•Improve patient outcomes
35. Why AI in healthcare now?
•The perfect storm
•Tons of data ("big data") generated by the minute
•Robust algorithms
•Processing power
36. Examples of AI in healthcare
• Patient self-monitoring (builds on the “quantified self”)
• Chronic and acute conditions; post-surgery
• PeerWell- AI for total joint replacement
• Patients begin using app before surgery. Patients receive customized daily
lessons and tasks which require them to input their results directly into the app.
Machine learning algorithm adjusts pre- and post-surgery instructions based on
patient input.
•Extended Visual Assistant (EVA)
• Voice-controlled eyewear for the visually impaired
• Uses machine learning to recognize objects, text, signs, etc, and verbally
describes what it sees
37. AI & Healthcare
• Patient self-monitoring, cont
• AI apps for more common ailments
• Diabetes management
• Palliative care
• Congenital heart disease
• Clinical trials
• Finding the right candidates for trials
• Predicting bioactivity of patients in trials
• Scheduling
• TrueCare’s Baymax bot
• Finds a doctor for you and then schedules an appointment
38. Examples of AI in Healthcare
• Pattern recognition
• Skin cancer detection
• Arterys- AI assistant for radiologists (1st FDA approval)
• Disease detection
39. How AI in Predictive Analytics/Modelling can benefit patients
• Increase the accuracy of diagnoses
• Improve preventive medicine and public health
• Enhance personalized care
• Accurately predict insurance costs
• Streamline research and development with prediction models
• Guide drug development to deliver medications that meet public need
• Better patient outcomes
40. AI in Research
•AIs can (and are!):
• Analyze data
• Clean data
• Identify patterns in data
• Discovery tools
• Gather data; e.g.; drone operators in places that are too
remote or dangerous for humans
• Analyze & match data; e.g., matching patients to clinical trials
41. Barriers to use of AI in healthcare
•Dirty data (data management has to happen
first)
•Silo’d data
•Discrimination
•Patients can’t be adequately served if
discriminated against
42. Barriers to use of AI in healthcare
•Lack of infrastructure/data management plan
•DMP should be predicated on International
Data Corporation (IDC) Third Platform
Principles, which are anchored by 4 areas:
•Big Data & Analytics
•Cloud
•Mobile
•Social
43. AI and Discrimination
• AI is only as unbiased as:
• the people who program it (mostly men; mostly white, in the US)
• the data on which its trained
• Deep neural networks for images most often trained on ImageNet,
which is heavily biased towards white Westerners
• Natural language processing algorithms trained on data sets scraped
from GoogleImages, GoogleNews, and Wikipedia- all heavily influenced
by men, which have led to misogynistic AIs
• “… Amazon’s computer models were trained to vet applicants by
observing patterns in resumes submitted to the company over a 10-year
period. Most came from men, a reflection of male dominance across the
tech industry.” – Amazon scraps secret AI recruiting tool that showed bias against
women (2018)
• the systems & institutions into which AI is deployed
44. AI & Discrimination: Training Humans
• “ first major photography exhibition devoted to training images: the
collections of photos used by scientists to train artificial intelligence (AI)
systems in how to ‘see’ and categorize the world.”
• Explores two fundamental issues:
• how humans are represented, interpreted and codified through
training datasets
• how technological systems harvest, label and use this material
45. AI & Discrimination: Training Humans
• Also of interest to the creators:
• Affect-training recognition, a system used by AI in security & hiring systems,
etc, to determine a person’s mental & emotional “aims to detect and
classify emotions by analyzing any face”
• ATR based on the work of Paul Ekman
• From his website: “Dr. Ekman identified the six basic emotions as anger,
surprise, disgust, enjoyment, fear, and sadness. His research shows the
strongest evidence to date of a seventh emotion, which is contempt.”
• Ekman’s work is highly disputed by other researchers in both the same
and different fields.
46. Affect- Training Recognition
• Now ATR is being used in hiring, security, and other systems to determine
whether someone would be a good employee or is a threat.
• HireVu
• “AI system to extract microexpressions, tone of voice, and other
variables from video job interviews, which it used to compare job
applicants against a company’s top performers.”
• Emotient (Apple)
• Claimed to be capable of identifying emotions from images of faces
• Affectiva
• detecting distracted and “risky” drivers on roads
• measuring consumers’ emotional responses to advertising
47. AI & Discrimination
• ImageNet Roulette- And, surprise(!), AI has some pretty racist and
misogynistic ideas about people. Or, rather, the dataset ImageNet
Roulette draws from, ImageNet, is filled with problematic
categories that reflect the bias often inherent in the large datasets
that make machine learning possible.
• ”Automated systems that replicate, and by extension exacerbate,
the biases present in society have the power to codify those very
problems.”
48. Future applications of AI in healthcare
• AI-powered predictive care
• networked hospitals
• better patient and staff experiences
50. Resources
• Timeline of Artificial Intelligence
https://en.wikipedia.org/wiki/Timeline_of_artificial_intelligence
• History of Machine Learning
https://www.estory.io/timeline/view/JlYn6L/445/History_of_Machine_Learning
• “What Is The Difference Between Artificial Intelligence And Machine
Learning?” https://bit.ly/2jHOxFA
• Everlaw- trial prep via document analysis https://www.everlaw.com/
• Voicera- https://www.voicera.com/
• Extended Visual Assistant (EVA), AI assistant
51. Resources
• “Neural Network | Human Brain versus computer”
https://techbuf.com/human-brain-neural-network/
• ShieldAI- drones for combat situations https://www.shield.ai/
• Wordsmith- https://automatedinsights.com/wordsmith
• AI’s role in healthcare starts small, gets bigger”
https://bit.ly/2F93JZu
• “How AI is transforming healthcare and solving problems in 2017”
https://bit.ly/2qWvugp
• “Google, Fitbit, startups storm into healthcare AI”
https://bit.ly/2ryvJgn
52. Resources
• “Artificial intelligence messenger bot, Baymax”
https://www.medicaldesignandoutsourcing.com/artificial-intelligence-messenger-bot/
• “’Big Hero 6': The Science Behind Baymax, Disney's Big, Soft Robot”
https://www.nbcnews.com/tech/gadgets/big-hero-6-science-behind-baymax-disneys-big-
soft-robot-n240241
• “No, Facebook Did Not Panic and Shut Down an AI Program That Was Getting Dangerously
Smart” https://gizmodo.com/no-facebook-did-not-panic-and-shut-down-an-ai-program-
1797414922
• “Google Self-Driving Cars Have Learned How to Interpret Cyclists' Hand Signals”
http://fortune.com/2016/07/06/google-self-driving-cars-cyclist/
• AI Chatbots Are Getting Better. But an Interview With ChatGPT Reveals Their Limits-
https://time.com/6238781/chatbot-chatgpt-ai-interview/
53. Resources
• “Predictive analytics in health care using machine learning tools and techniques”
https://ieeexplore.ieee.org/document/8250771/
• “How artificial intelligence is revolutionizing the patient experience in healthcare”
https://www.telusinternational.com/articles/ai-patient-experience-healthcare/
• “'It Is Crazy!' The Promise and Potential Peril of ChatGPT”
https://www.medpagetoday.com/opinion/patientcenteredmedicalhome/102557
• Creating Artificial Intelligence 'In Full Color’
https://www.nursing.virginia.edu/news/ai-ecosystem-williams-moorman/
• “These ER Docs Invented a Real Star Trek Tricorder”
https://www.nbcnews.com/mach/technology/these-er-docs-invented-real-star-trek-
tricorder-n755631
• “What Companies Are Winning The Race For Artificial Intelligence?”
https://www.forbes.com/sites/quora/2017/02/24/what-companies-are-winning-the-
race-for-artificial-intelligence/#34820637f5cd
54. Resources
• “Just a Few of the Amazing Things AI Is Doing in Healthcare”
https://singularityhub.com/2018/03/29/just-a-few-of-the-amazing-things-ai-is-doing-in-
healthcare/#sm.00000ffrfb4hgpe2xxzxbtpckn6ws
• “Artificial intelligence powers digital medicine” https://www.nature.com/articles/s41746-017-0012-2
• “Man against machine: AI is better than dermatologists at diagnosing skin cancer”
https://www.eurekalert.org/pub_releases/2018-05/esfm-mam052418.php
• "Contributed: Top 10 Use Cases for AI in Healthcare”
https://www.mobihealthnews.com/news/contributed-top-10-use-cases-ai-healthcare
• Can Artificial Intelligence detect Melanoma?
https://www.mskcc.org/news/can-artificial-intelligence-detect-melanoma
• Artificial intelligence in disease diagnosis: a systematic literature review, synthesizing framework and
future research agenda- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8754556/
55. Resources
• What Do We Do About the Biases in AI?- https://hbr.org/2019/10/what-do-we-
do-about-the-biases-in-ai
• How tech's white male workforce feeds bias into AI-
https://www.cbsnews.com/news/ai-bias-problem-techs-white-male-workforce/
• AI can be sexist and racist — it’s time to make it fair-
https://www.nature.com/articles/d41586-018-05707-8
• Amazon scraps secret AI recruiting tool that showed bias against women-
https://www.reuters.com/article/us-amazon-com-jobs-automation-
insight/amazon-scraps-secret-ai-recruiting-tool-that-showed-bias-against-
women-idUSKCN1MK08G
•
56. Resources
• AINOW 2019 Report- https://ainowinstitute.org/AI_Now_2019_Report.pdf
• ChatGPT: five priorities for research- https://www.nature.com/articles/d41586-023-
00288-7
• Replika AI- https://replika.com/
• Project December- https://projectdecember.net/
• Hereafter AI- https://www.hereafter.ai/
• Good Bot, Bad Bot | Part III: Life, death and AI- Endless Thread podcast-
https://www.wbur.org/endlessthread/2022/11/18/life-death-ai
• Good Bot, Bad Bot | Part I: Mental Health and Bot Therapy- Endless Thread podcast-
https://www.wbur.org/endlessthread/2022/11/04/bots-mental-health
• Whispers of AI’s Modular Future- https://www.newyorker.com/tech/annals-of-
technology/whispers-of-ais-modular-future
57. Resources
• Arterys Resources https://www.arterys.com/resources-library
• “The future is now? Zipline http://www.flyzipline.com/
• “Zipline, which delivers lifesaving medical supplies by drone, now valued at $1.2 billion”
https://www.cnbc.com/2019/05/17/zipline-medical-delivery-drone-start-up-now-valued-at-
1point2-billion.html
• AITopics https://aitopics.org/search
• Nonhuman “Authors” and Implications for the Integrity of Scientific Publication and Medical
Knowledge https://jamanetwork.com/journals/jama/fullarticle/2801170
• Peer-Reviewed Journal Publishes Paper Written Almost Entirely by ChatGPT
https://www.medpagetoday.com/special-reports/exclusives/102960?
• Worldwide Artificial Intelligence Spending Guide-
https://www.idc.com/getdoc.jsp?containerId=IDC_P33198
58. Resources
• Google Bard AI hands-on: A work in progress with plenty of caveats-
https://www.engadget.com/google-bard-ai-hands-on-a-work-in-progress-with-plenty-of-
caveats-170956025.html
• Microsoft Adds DALL-E AI Image Generator to Bing- https://gizmodo.com/microsoft-adds-
dall-e-ai-image-generator-to-bing-1850247593
• Microsoft's new AI chatbot has been saying some 'crazy and unhinged things’-
https://www.npr.org/2023/03/02/1159895892/ai-microsoft-bing-chatbot
• See what AI really thinks of you with this deeply humbling website-
https://mashable.com/article/ai-machine-learning-imagenet-roulette
59. Resources
• Artificial Intelligence is Misreading Human Emotion-
https://www.theatlantic.com/technology/archive/2021/04/artificial-intelligence-misreading-
human-emotion/618696/
• Global Emotion Detection & Recognition Market Size is Projected to Grow from USD 21.6
Billion in 2019 to USD 56.0 Billion by 2024, at a CAGR of 21.0% - ResearchAndMarkets.com-
https://www.businesswire.com/news/home/20200213005614/en/Global-Emotion-Detection-
Recognition-Market-Size-is-Projected-to-Grow-from-USD-21.6-Billion-in-2019-to-USD-56.0-
Billion-by-2024-at-a-CAGR-of-21.0---ResearchAndMarkets.com
• World Economic Forum: 3 ways AI will transform healthcare in the next decade-
https://www.beckershospitalreview.com/innovation/world-economic-forum-3-ways-ai-will-
transform-healthcare-in-the-next-decade?oly_enc_id=6411D3520189C0K